X-ray Structure of the Ca2+-binding Interaction Domain of C1s
2003; Elsevier BV; Volume: 278; Issue: 34 Linguagem: Inglês
10.1074/jbc.m305175200
ISSN1083-351X
AutoresLynn Gregory, Nicole M. Thielens, Gérard J. Arlaud, Juan C. Fontecilla‐Camps, Christine Gaboriaud,
Tópico(s)Blood Coagulation and Thrombosis Mechanisms
ResumoC1, the complex that triggers the classical pathway of complement, is assembled from two modular proteases C1r and C1s and a recognition protein C1q. The N-terminal CUB1-EGF segments of C1r and C1s are key elements of the C1 architecture, because they mediate both Ca2+-dependent C1r-C1s association and interaction with C1q. The crystal structure of the interaction domain of C1s has been solved and refined to 1.5 Å resolution. The structure reveals a head-to-tail homodimer involving interactions between the CUB1 module of one monomer and the epidermal growth factor (EGF) module of its counterpart. A Ca2+ ion is bound to each EGF module and stabilizes both the intra- and inter-monomer interfaces. Unexpectedly, a second Ca2+ ion is bound to the distal end of each CUB1 module, through six ligands contributed by Glu45, Asp53, Asp98, and two water molecules. These acidic residues and Tyr17 are conserved in approximately two-thirds of the CUB repertoire and define a novel, Ca2+-binding CUB module subset. The C1s structure was used to build a model of the C1r-C1s CUB1-EGF heterodimer, which in C1 connects C1r to C1s and mediates interaction with C1q. A structural model of the C1q/C1r/C1s interface is proposed, where the rod-like collagen triple helix of C1q is accommodated into a groove along the transversal axis of the C1r-C1s heterodimer. C1, the complex that triggers the classical pathway of complement, is assembled from two modular proteases C1r and C1s and a recognition protein C1q. The N-terminal CUB1-EGF segments of C1r and C1s are key elements of the C1 architecture, because they mediate both Ca2+-dependent C1r-C1s association and interaction with C1q. The crystal structure of the interaction domain of C1s has been solved and refined to 1.5 Å resolution. The structure reveals a head-to-tail homodimer involving interactions between the CUB1 module of one monomer and the epidermal growth factor (EGF) module of its counterpart. A Ca2+ ion is bound to each EGF module and stabilizes both the intra- and inter-monomer interfaces. Unexpectedly, a second Ca2+ ion is bound to the distal end of each CUB1 module, through six ligands contributed by Glu45, Asp53, Asp98, and two water molecules. These acidic residues and Tyr17 are conserved in approximately two-thirds of the CUB repertoire and define a novel, Ca2+-binding CUB module subset. The C1s structure was used to build a model of the C1r-C1s CUB1-EGF heterodimer, which in C1 connects C1r to C1s and mediates interaction with C1q. A structural model of the C1q/C1r/C1s interface is proposed, where the rod-like collagen triple helix of C1q is accommodated into a groove along the transversal axis of the C1r-C1s heterodimer. The classical pathway of complement, a major element of innate immunity against pathogens, is triggered by C1, a 790-kDa complex formed from association of a recognition protein C1q with two modular serine proteases, C1r and C1s, that respectively mediate internal activation and proteolytic activity of the complex (1Cooper N.R. Adv. Immunol. 1985; 37: 151-216Crossref PubMed Scopus (391) Google Scholar, 2Arlaud G.J. Colomb M.G. Gagnon J. Immunol. Today. 1987; 8: 106-111Abstract Full Text PDF PubMed Scopus (80) Google Scholar, 3Arlaud G.J. Gaboriaud C. Thielens N.M. Rossi V. Bersch B. Hernandez J.-F. Fontecilla-Camps J.C. Immunol. Rev. 2001; 180: 136-145Crossref PubMed Scopus (66) Google Scholar). C1q is a protein with the overall shape of a bouquet of flowers, comprising six heterotrimeric collagen-like triple helices that associate to form a N-terminal "stalk" and then diverge to form individual "stems," each terminating in a C-terminal globular domain (Ref. 4Kishore U. Reid K.B.M. Immunopharmacology. 2000; 49: 159-170Crossref PubMed Scopus (393) Google Scholar; see Fig. 5). C1r and C1s have homologous modular structures with a N-terminal C1r/C1s, uEGF, bone morphogenetic protein (CUB) module (5Bork P. Beckmann G. J. Mol. Biol. 1993; 231: 539-545Crossref PubMed Scopus (521) Google Scholar), an epidermal growth factor (EGF)-like 1The abbreviations used are: EGF, epidermal growth factor; MASP, mannan-binding lectin-associated serine protease; ESRF, European Synchrotron Radiation Facility. module of the Ca2+-binding type (6Campbell I.D. Bork P. Curr. Opin. Struct. Biol. 1993; 3: 385-392Crossref Scopus (331) Google Scholar), a second CUB module, two complement control protein modules (7Reid K.B.M. Bentley D.R. Campbell R.D. Chung L.P. Sim R.B. Kristensen T. Tack B.F. Immunol. Today. 1986; 7: 230-234Abstract Full Text PDF PubMed Scopus (187) Google Scholar), and a chymotrypsin-like serine protease domain. This modular architecture is shared by the mannan-binding lectin-associated serine proteases (MASPs), a group of enzymes involved in the triggering of the lectin pathway of complement (8Fujita T. Nat. Rev. Immunol. 2002; 2: 346-353Crossref PubMed Scopus (569) Google Scholar). Whereas the enzymatic properties of C1r and C1s are mediated by their C-terminal regions, the N-terminal CUB1-EGF domains have interaction properties that are essential to the assembly of the C1 complex. Thus, it is well established that C1s-C1r-C1r-C1s, the tetrameric catalytic subunit of C1, assembles through Ca2+-dependent heterodimeric C1r-C1s interactions involving the CUB1-EGF segment of each protease (9Thielens N.M. Aude C.A. Lacroix M.B. Gagnon J. Arlaud G.J. J. Biol. Chem. 1990; 265: 14469-14475Abstract Full Text PDF PubMed Google Scholar, 10Busby T.F. Ingham K.C. Biochemistry. 1990; 29: 4613-4618Crossref PubMed Scopus (58) Google Scholar, 11Tsai S.-W. Poon P.H. Schumaker V.N. Mol. Immunol. 1997; 34: 1273-1280Crossref PubMed Scopus (17) Google Scholar, 12Thielens N.M. Enrié K. Lacroix M. Jaquinod M. Hernandez J.-F. Esser A.F. Arlaud G.J. J. Biol. Chem. 1999; 274: 9149-9159Abstract Full Text Full Text PDF PubMed Scopus (56) Google Scholar). Furthermore, the current data (10Busby T.F. Ingham K.C. Biochemistry. 1990; 29: 4613-4618Crossref PubMed Scopus (58) Google Scholar, 11Tsai S.-W. Poon P.H. Schumaker V.N. Mol. Immunol. 1997; 34: 1273-1280Crossref PubMed Scopus (17) Google Scholar, 13Thielens N.M. Illy C. Bally I.M. Arlaud G.J. Biochem. J. 1994; 301: 378-384Crossref Scopus (28) Google Scholar) are consistent with the hypothesis that the CUB1-EGF moieties of C1r and C1s each contribute ligands for the interaction between the C1s-C1r-C1r-C1s tetramer and C1q sites located in the individual collagen-like stems of the protein (14Siegel R.C. Schumaker V.N. Mol. Immunol. 1983; 20: 53-66Crossref PubMed Scopus (65) Google Scholar, 15Strang C.J. Siegel R.C. Philipps M.L. Poon P.H. Schumaker V.N. Proc. Natl. Acad. Sci. U. S. A. 1982; 79: 586-590Crossref PubMed Scopus (72) Google Scholar). Based on these and other features, several low resolution models of the C1 complex have been proposed (2Arlaud G.J. Colomb M.G. Gagnon J. Immunol. Today. 1987; 8: 106-111Abstract Full Text PDF PubMed Scopus (80) Google Scholar, 16Schumaker V.N. Hanson D.C. Kilchherr E. Philipps M.L. Poon P.H. Mol. Immunol. 1986; 23: 557-565Crossref PubMed Scopus (51) Google Scholar, 17Weiss V. Fauser C. Engel J. J. Mol. Biol. 1986; 189: 573-581Crossref PubMed Scopus (48) Google Scholar). In an effort to decipher the structure-function relationships of the C1 complex at the atomic level, we have used a dissection strategy that has yielded precise insights into the activation mechanism of C1r (18Budayova-Spano M. Lacroix M. Thielens N.M. Arlaud G.J. Fontecilla-Camps J.C. Gaboriaud C. EMBO J. 2002; 21: 231-239Crossref PubMed Scopus (91) Google Scholar, 19Budayova-Spano M. Grabarse W. Thielens N.M. Hillen H. Lacroix M. Schmidt M. Fontecilla-Camps J.C. Arlaud G.J. Gaboriaud C. Structure. 2002; 10: 1509-1519Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar) and the proteolytic function of C1s (20Gaboriaud C. Rossi V. Bally I. Arlaud G.J. Fontecilla-Camps J.C. EMBO J. 2000; 19: 1755-1765Crossref PubMed Scopus (93) Google Scholar). We now report the x-ray structure of the CUB1-EGF moiety of C1s, a domain that in the C1 complex associates with the corresponding part of C1r and has the additional ability to form homodimers in the absence of C1r (9Thielens N.M. Aude C.A. Lacroix M.B. Gagnon J. Arlaud G.J. J. Biol. Chem. 1990; 265: 14469-14475Abstract Full Text PDF PubMed Google Scholar, 10Busby T.F. Ingham K.C. Biochemistry. 1990; 29: 4613-4618Crossref PubMed Scopus (58) Google Scholar, 11Tsai S.-W. Poon P.H. Schumaker V.N. Mol. Immunol. 1997; 34: 1273-1280Crossref PubMed Scopus (17) Google Scholar). The structure reveals a novel, Ca2+-binding CUB module subset and yields insights into the C1q/C1r/C1s interface in the C1 complex. Production and Purification of the Recombinant C1s CUB1-EGF Domain—A DNA fragment encoding the C1s signal peptide and the N-terminal CUB1-EGF segment (residues 1–159 of the mature protein) was amplified by PCR using VentR polymerase and the pBS-C1s plasmid (21Luo C. Thielens N.M. Gagnon J. Gal P. Sarvari M. Tseng Y. Tosi M. Zavodszky P. Arlaud G.J. Schumaker V.N. Biochemistry. 1992; 31: 4254-4262Crossref PubMed Scopus (35) Google Scholar) as a template, according to established procedures. The sequences of the sense (5′-CGGGATCCATGTGGTGCATTGTCCTG-3′) and antisense (5′-GGGGTACC CTAATTAACTCCGCAATTCTTC-3′) primers introduced a BamHI restriction site (underlined) at the 5′ end of the polymerase chain reaction product and a stop codon (bold type) followed by a KpnI site (underlined) at the 3′ end. The amplified DNA was purified using the Geneclean kit (Bio 101), digested with BamHI and KpnI, and cloned into the corresponding sites of the pFastBac1 baculovirus transfer vector (Invitrogen). The resulting construct was characterized by restriction mapping and checked by double-stranded DNA sequencing (Genome Express, Grenoble, France). The recombinant baculovirus was generated using the Bac-to-Bac™ system (Invitrogen Corp.), amplified, and titrated as described previously (22Thielens N.M. Cseh S. Thiel S. Vorup-Jensen T. Rossi V. Jensenius J.C. Arlaud G.J. J. Immunol. 2001; 166: 5068-5077Crossref PubMed Scopus (114) Google Scholar). High Five cells (1.75 × 107 cells/175-cm2 tissue culture flask) were infected with the recombinant virus at a multiplicity of infection of 2 in Sf900 II SFM medium (Invitrogen) for 96 h at 28 °C. The supernatant was collected by centrifugation, and diisopropyl phosphorofluoridate was added to a final concentration of 1 mm. The culture supernatant containing the C1s CUB1-EGF segment was dialyzed against 75 mm NaCl, 10 mm imidazole, pH 6.1, and loaded at 1.5 ml/min onto a Q-Sepharose-Fast Flow column (Amersham Biosciences) (2.8 × 12 cm) equilibrated in the same buffer. Elution was carried out by applying a 1-liter linear gradient from 75 to 500 mm NaCl in the same buffer. The fractions containing the recombinant fragment were identified by SDS-PAGE analysis, precipitated by addition of (NH4)2SO4 to 60% (w/v), and left overnight at 4 °C. The pellets were resuspended in 145 mm NaCl, 1 mm EDTA, 50 mm triethanolamine hydrochloride, pH 7.4, and applied onto a TSK G3000 SWG column (7.5 × 600 mm) (Toso Haas) equilibrated in the same buffer. The purified fragment was concentrated to 1.0 mg/ml by ultrafiltration on a PM-10 membrane (Amicon). Chemical and Functional Characterization of the Recombinant Protein—SDS-PAGE analysis was performed as described previously (9Thielens N.M. Aude C.A. Lacroix M.B. Gagnon J. Arlaud G.J. J. Biol. Chem. 1990; 265: 14469-14475Abstract Full Text PDF PubMed Google Scholar). Mass spectrometry analysis was performed using the matrix-assisted laser desorption ionization technique on a Voyager Elite XL instrument (PerSeptive Biosystems, Cambridge, MA) under conditions described previously (23Lacroix M. Rossi V. Gaboriaud C. Chevallier S. Jaquinod M. Thielens N.M. Gagnon J. Arlaud G.J. Biochemistry. 1997; 36: 6270-6282Crossref PubMed Scopus (43) Google Scholar). High pressure gel permeation chromatography was performed on a TSK G3000 SWG column (7.5 × 600 mm) (Tosohaas) equilibrated in 145 mm NaCl, 50 mm triethanolamine hydrochloride, pH 7.4, containing either 1 mm EDTA or CaCl2 and run at 1 ml/min. Analysis by surface plasmon resonance spectroscopy of the interaction between the C1s CUB1-EGF domain and intact C1r was performed at 25 °C using an upgraded BIAcore instrument (BIAcore AB, Uppsala, Sweden). The running buffer for protein immobilization was 145 mm NaCl, 5 mm EDTA, 10 mm HEPES, pH 7.4. The C1s CUB1-EGF domain was diluted to 35 μg/ml in 10 mm formate, pH 3.0, and coupled to the carboxymethylated dextran surface of a CM5 sensor chip (BIAcore AB) using the amine coupling chemistry (BIAcore amine coupling kit). Binding of purified plasma-derived human C1r (24Arlaud G.J. Sim R.B. Duplaa A.-M. Colomb M.G. Mol. Immunol. 1979; 16: 445-450Crossref PubMed Scopus (112) Google Scholar) was measured over 250 resonance units of the immobilized C1s CUB1-EGF segment, at a flow rate of 10 μl/min in 145 mm NaCl, 1 mm CaCl2, 50 mm triethanolamine hydrochloride, pH 7.4. Equivalent volumes of the C1r samples were injected over a surface with immobilized ovalbumin to serve as blank sensorgrams for subtraction of bulk refractive index background. Regeneration of the surface was achieved by injection of 10 μl of 5 mm EDTA. The data were analyzed by global fitting to a 1:1 Langmuir binding model of both the association and dissociation phases for several concentrations simultaneously, using the BIAevaluation 3.1 software (BIAcore). The apparent equilibrium dissociation constant (K D) was calculated from the ratio of the dissociation and association rate constants (k off/k on). Crystallization and Data Collection—The C1s CUB1-EGF fragment was concentrated to 6.0–7.8 mg/ml in 145 mm NaCl, 1 mm CaCl2,50mm triethanolamine HCl, pH 7.4. The crystals were obtained at 20 °Cbythe hanging drop vapor diffusion method by mixing equal volumes of the protein solution and of a reservoir solution composed of 30% (v/v) PEG 400, 0.2 m CaCl2, and 0.1 m HEPES, pH 7.5. The crystals obtained were used for micro-seeding in a drop with reservoir solution containing 16% (v/w) PEG 4000, 0.2 m MgCl2, 8.7% glycerol, and 0.1 m HEPES, pH 7.5. From this were obtained crystals suitable for high resolution x-ray data collection. A native data set with space group P1 was measured at the ESRF beamline ID14-EH2 to a resolution of 1.5 Å. The images were processed using the MOSFLM program package (25Leslie A.G.W. Moras D. Podjarny A.D. Thierry J.C. Crystallographic Computing. Oxford University Press, Oxford1991: 50-61Google Scholar), and the data were scaled using the Collaborative Computational Project 4 suite (26Collaborative Computational Project 4Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19770) Google Scholar). Details are given in Table I.Table IData collection and refinement statisticsNativeaCollected at ID14-eh2, ESRF.SADbCollected at BM30 FIP, ESRF.Space groupP1P1Unit cell (Å)a = 35.14 b = 47.50 c = 56.68a = 35.38 b = 47.31 c = 57.40(°)α = 87.74 β = 78.04 γ = 75.67α = 87.58 β = 79.76 γ = 75.81λ (Å)0.9331.649 ÅResolution (Å)30-1.5035-3.00R sym0.064 (0.143)cStatistics for high resolution bin (1.58-1.50 Å) are in parentheses.0.096 (0.191)dStatistics for high resolution bin (3.0-3.15 Å) are in parentheses.% completeness94.4 (93.3)94.5 (88.6)Redundancy3.5 (2.2)2.98 (2.21)I/sigma (I) average6.9 (4.0)13.02 (6.35)No. of reflections181271 (16763)39958 (3794)No. of unique reflections52518 (7588)13394 (1713)Figure of merit0.4362Phasing power2.4539Model StatisticsFinal resolution (Å)1.5No. of residues311No. of water molecules283No. of ions4Root mean square deviation χ2 bonds (Å)0.0064Root mean square deviation χ2 angles (°)1.43R work0.229R free0.246a Collected at ID14-eh2, ESRF.b Collected at BM30 FIP, ESRF.c Statistics for high resolution bin (1.58-1.50 Å) are in parentheses.d Statistics for high resolution bin (3.0-3.15 Å) are in parentheses. Open table in a new tab Structure Determination and Refinement—A preliminary structure was solved using the single-wavelength anomalous dispersion method. Heavy atom derivatives were prepared by first transferring the crystal to a solution with a lower concentration of MgCl2 (0.15 m) and then by soaking in a mother liquor containing 0.5 mm TbCl3 and 0.15 m MgCl2. The heavy atom derivative data set was collected at the ESRF beamline BM30 at 3.0 Å resolution and indexed using XDS (27Kabsch W. Rossmann M.G. Arnold E. International Tables for Crystallography. Kluwer Academic Publishers, Dordrecht, The Netherlands2001: 730-734Google Scholar). Two heavy derivative sites were located using the Patterson heavy atom search method implemented in CNS (28Brünger A.T. Adams P.D. Clore G.M. DeLano W.L. Gros P. Grosse-Kunstleve R.W. Jiang J.S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice L.M. Simonson T. Warren G.L. Acta Crystallogr. Sect. D Biol. Crystallogr. 1998; 54: 905-921Crossref PubMed Scopus (16967) Google Scholar). The correct enantiomorph was found by visual inspection of the electron density map. The protein region was distinguishable from the solvent. Solvent flattening was carried out with DM (29Cowtan K. Main P. Acta Crystallogr. Sect. D Biol. Crystallogr. 1998; 54: 487-493Crossref PubMed Scopus (309) Google Scholar). Model building was carried out with the graphics program TURBO (30Roussel A. Cambillau C. Silicon Graphics Geometry Partners Directory. Silicon Graphics, Mountain View, CA1991: 86Google Scholar). The quality of the initial maps permitted the construction of up to 60% of the 318 residues in the asymmetric unit. This initial model was refined using CNS and then used as the molecular replacement search model for the high resolution native data set. The ensuing map was improved with WARP, v5.0 (31Perrakis A. Morris R. Lamzin V.S. Nat. Struct. Biol. 1999; 5: 458-463Crossref Scopus (2564) Google Scholar), allowing automatic building of 86% of the model (269 residues) at this stage. The automatic water molecule search and refinement for the high resolution model was done with CNS. The atomic coordinates have been deposited in the Protein Data Bank under the code 1NZI. Modeling of the C1r/C1s/C1q Interface—Modeling of the C1r-C1s CUB1-EGF heterodimer was carried out in three steps. First, the structure of the C1s CUB1 module was used as a scaffold for the C1r CUB1 model. After careful examination of the sequence alignment of the CUB1 modules of C1s and C1r, the residues in the structure of C1s CUB1 were replaced by the corresponding residues from the C1r sequence using the graphics program TURBO (30Roussel A. Cambillau C. Silicon Graphics Geometry Partners Directory. Silicon Graphics, Mountain View, CA1991: 86Google Scholar). Next, the C1r EGF structure determined by NMR spectroscopy (32Bersch B. Hernandez J.-F. Marion D. Arlaud G.J. Biochemistry. 1998; 37: 1204-1214Crossref PubMed Scopus (44) Google Scholar) was superimposed onto the atomic coordinates of the C1s EGF B module using the interactive graphics program O (33Jones T.A. Zou J.Y. Lowan S.W. Kjeldgaard M. Acta Crystallogr. Sect. A. 1991; 47: 110-119Crossref PubMed Scopus (13011) Google Scholar). The program gave an root mean square of 1.05 Å based on the 33 Cα positions used to carry out the superimposition. Loop 10 of each EGF module was excluded from the this calculation because of the difference in length and conformation. Finally, the heterodimer was assembled by taking the Protein Data Bank file of the C1s homodimer and superimposing the C1r CUB1 and EGF models onto monomer B of the C1s homodimer. The model of the C1q collagen arm is based on published statistical information derived from collagen-like structures (34Rainey J.K. Goh M.C. Protein Sci. 2002; 11: 2748-2754Crossref PubMed Scopus (60) Google Scholar). The arrangement of the A, B, and C chains in the heterotrimeric triple helix is derived from the crystal structure of the C1q globular domain, 2C. Gaboriaud, J. Juanhuix, A. Gruez, M. Lacroix, C. Darnault, D. Pignol, D. Verger, J-C. Fontecilla-Camps, and G. Arlaud, unpublished data. and the alignment of the collagen triplets shown in Fig. 5 is the only one compatible with this structure. Different configurations of the C1q/C1r/C1s interface were tested using computer graphics, looking for the most appropriate positioning in terms of shape and charge complementarity between the C1q and C1r-C1s models. Consistency with previously published biochemical data was also included in the selection of the most plausible model (see "Discussion"). The model depicted in Fig. 5 allows ionic interaction between unmodified lysine residues A59, B61, and B65 of C1q and acidic residues of C1r (Asp61, Glu137 or Glu138, and Asp127, respectively). In this configuration, hydrophobic residues of C1q are positioned in a favorable environment at the C1s-C1r interface: methionines B68 and C67 point toward the central six-residue hydrophobic cluster, whereas residues A74, A77, C70, C71, and C74 are in the vicinity of the distal hydrophobic pocket. Characterization of the C1s Interaction Domain—Expression of the C1s CUB1-EGF domain in a baculovirus/insect cells system led to the production of large amounts of material (∼20 mg/liter of cell culture). Purification was achieved by ion-exchange chromatography followed by (NH4)2SO4 precipitation and gel permeation. SDS-PAGE analysis showed that the purified protein was homogeneous and migrated as a single band with an apparent molecular mass of 25.5 and 20.5 kDa under reducing and nonreducing conditions, respectively (Fig. 1). Mass spectrometry analysis yielded a value of 18,129 ± 9 Da consistent with the amino acid sequence of the N-terminal Glu1–Asn159 segment of C1s (calculated mass, 18,125 Da). Analysis by surface plasmon resonance spectroscopy showed that the immobilized C1s CUB1-EGF domain was able to bind C1r in a Ca2+-dependent fashion, with a K D value of 20.7 nm, similar to the values of 10.9 and 20.2 nm determined previously for intact C1s and the larger N-terminal C1sα fragment, respectively (12Thielens N.M. Enrié K. Lacroix M. Jaquinod M. Hernandez J.-F. Esser A.F. Arlaud G.J. J. Biol. Chem. 1999; 274: 9149-9159Abstract Full Text Full Text PDF PubMed Scopus (56) Google Scholar). Gel filtration analysis of the C1s CUB1-EGF domain indicated that the protein eluted significantly earlier in the presence of Ca2+ ions than in the presence of EDTA, consistent the known ability of the C1s interaction domain to form Ca2+-dependent homodimers (9Thielens N.M. Aude C.A. Lacroix M.B. Gagnon J. Arlaud G.J. J. Biol. Chem. 1990; 265: 14469-14475Abstract Full Text PDF PubMed Google Scholar, 10Busby T.F. Ingham K.C. Biochemistry. 1990; 29: 4613-4618Crossref PubMed Scopus (58) Google Scholar, 11Tsai S.-W. Poon P.H. Schumaker V.N. Mol. Immunol. 1997; 34: 1273-1280Crossref PubMed Scopus (17) Google Scholar). Overall Structure—The structure of the CUB1-EGF interaction domain of human C1s was solved by the single-wavelength anomalous dispersion method and refined at 1.5 Å into a very well defined electron density map (Fig. 2D). The final R work and R free factors are 0.229 and 0.246, respectively, and the refined model has excellent stereochemistry (Table I). In agreement with previous findings (9Thielens N.M. Aude C.A. Lacroix M.B. Gagnon J. Arlaud G.J. J. Biol. Chem. 1990; 265: 14469-14475Abstract Full Text PDF PubMed Google Scholar, 10Busby T.F. Ingham K.C. Biochemistry. 1990; 29: 4613-4618Crossref PubMed Scopus (58) Google Scholar, 11Tsai S.-W. Poon P.H. Schumaker V.N. Mol. Immunol. 1997; 34: 1273-1280Crossref PubMed Scopus (17) Google Scholar), the CUB1-EGF segment of C1s associates as a Ca2+-dependent homodimer (Fig. 2). Within each monomer, the CUB1 and EGF modules are assembled in a linear fashion with a Ca2+ ion bound at the intermodular interface (site I). A second Ca2+ ion is bound to the distal part of each CUB1 module (site II). The two monomers interact in a head-to-tail fashion involving major contacts between the CUB1 module of one molecule and the EGF module of its counterpart, the resulting assembly displaying a noncrystallographic pseudo 2-fold symmetry. The C-terminal residues are located at either end of the dimer, indicating where the CUB2 modules follow (Fig. 2, A and B). The overall structure is rather elongated, with a length of approximately 85 Å and a width of 20–40 Å. A side view of the structure (Fig. 2C) reveals that whereas one side is relatively flat, the opposite side is markedly concave and forms a groove. A Novel, Ca2 + -binding CUB Module Structure—Compared with the CUB domain topology established from the x-ray structure of two spermadhesins (35Romero A. Romao J.J. Varela P.F. Kölln I. Dias J.M. Carvalho A.L. Sanz L. Topfer-Petersen E. Calvete J.J. Nat. Struct. Biol. 1997; 4: 783-788Crossref PubMed Scopus (129) Google Scholar), the C1s CUB1 module reveals a number of particular features (Fig. 3). Like the N-terminal CUB module in C1r and the MASPs, the C1s CUB1 module shows a deletion at its N-terminal end (Fig. 3C). As a result, this module lacks not only the first of the two disulfide bridges characteristic of most CUB domains but also the first two β-strands present in the previously determined CUB structures (Fig. 3A). Thus, whereas CUB domains of the spermadhesin family are organized in two five-stranded β-sheets, each containing two parallel and four anti-parallel strands (35Romero A. Romao J.J. Varela P.F. Kölln I. Dias J.M. Carvalho A.L. Sanz L. Topfer-Petersen E. Calvete J.J. Nat. Struct. Biol. 1997; 4: 783-788Crossref PubMed Scopus (129) Google Scholar), the C1s CUB1 topology consists of two four-stranded β-sheets, each made of anti-parallel strands (strands 3, 10, 5, and 8 and strands 4, 9, 6, and 7). A further specific feature of the C1s CUB1 structure is the 3/10 helical conformation of the loop (H1) connecting strands β5 to β6, which is deleted in the spermadhesins (Fig. 3, A and C). Loops 3 and 9, on the same side of the module, and the large insertion loop 7 (a specific feature of the C1r/C1s/MASP family) also exhibit significant differences in length and/or conformation compared with their counterparts in the spermadhesin CUB structures (Figs. 3, A and C). In contrast to these modifications in solvent-exposed regions, the hydrophobic core observed in the spermadhesin family is highly conserved in the C1s CUB1 domain, the 18 hydrophobic or aromatic residues defining the CUB domain signature (5Bork P. Beckmann G. J. Mol. Biol. 1993; 231: 539-545Crossref PubMed Scopus (521) Google Scholar) being conserved in C1s (Fig. 3C). Compared with the spermadhesins, the C1s CUB1 domain shows root mean square deviation values of 1.40 Å (aSFP), 1.50 Å (PSP-I), and 1.54 Å (PSP-II), based on 80–86 homologous residues. An unexpected feature of the structure is the occurrence of a Ca2+-binding site (site II) at the distal end of each CUB1 module. The Ca2+ ion is coordinated by six oxygen ligands, namely one side chain oxygen of Glu45, both carboxylate oxygens of Asp53, the main chain carbonyl oxygen of Asp98, and two water molecules (Fig. 3B). The bond distances are in average 2.4 Å, the characteristic value for known Ca2+-binding sites in proteins (36Harding M. Acta Crystallogr. Sect. D Biol. Crystallogr. 1999; 55: 1432-1443Crossref PubMed Scopus (234) Google Scholar). In addition, the Ca2+ ion, its ligands, and the neighboring residues Tyr17, Asn101, and Phe105 also partake in an intricate network of hydrogen bonds that connect together strands β5, β6, β9, and β10 and loops L3 and L9 (Fig. 3B). Thus, the Ca2+ ion is the central element of a network of interactions that extensively stabilize the distal end of the C1s CUB1 module. The Ca2+ ion in site II is exposed to the solvent and exchangeable for Tb3+, as seen in the heavy atom derivative. Partial replacement by Mg2+ was also observed in crystals grown at high MgCl2 concentrations. Subtle differences in the coordinating ligands were observed between monomers A and B of the homodimer, including in some cases the involvement of a further water molecule contributing a seventh ligand. Of the residues involved in the coordination of Ca2+ and the associated network of hydrogen bonds, Asn101 and Phe105 appear to be strictly specific to the CUB1 modules of the C1r/C1s/MASP family (Fig. 3B). In contrast, Tyr17 and the Ca2+ ligands Glu45, Asp53, and Asp98 are conserved in approximately two-thirds of the CUB module repertoire, strongly suggesting that these residues define a novel CUB module subset with the ability to bind Ca2+. This subset is possibly more representative in terms of structure than the spermadhesins. Ca2 + -binding Site I and the Intra-monomer CUB1-EGF Interface—The C1s EGF module exhibits a fold similar to that described for other modules of this type (6Campbell I.D. Bork P. Curr. Opin. Struct. Biol. 1993; 3: 385-392Crossref Scopus (331) Google Scholar), with one major and one minor anti-parallel double-stranded β-sheets (Fig. 2A). Loop 10, which is disordered in the C1r EGF module (32Bersch B. Hernandez J.-F. Marion D. Arlaud G.J. Biochemistry. 1998; 37: 1204-1214Crossref PubMed Scopus (44) Google Scholar), is much shorter in C1s (Fig. 4B) and structurally well defined, except Phe123 in monomer A. The remainder of the C1r and C1s EGF modules shows a root mean square deviation value of 0.90 Å. As predicted from the amino acid sequence (3Arlaud G.J. Gaboriaud C. Thielens N.M. Rossi V. Bersch B. Hernandez J.-F. Fontecilla-Camps J.C. Immunol. Rev. 2001; 180: 136-145Crossref PubMed Scopus (66) Google Scholar, 6Campbell I.D. Bork P. Curr. Opin. Struct. Biol. 1993; 3: 385-392Crossref Scopus (331) Google Scholar), a Ca2+ ion (site I) is bound to both EGF modules of the CUB1-EGF homodimer (Fig. 2A). The Ca2+ ion is coordinated by seven oxygen ligands, including a wate
Referência(s)